Model Repair in Systems Design Panagiotis Katsaros Aristotle University of Thessaloniki (GR)
Model-Based Design for Space Systems @ AUTh Design Validation Studies Using COMPASS! Bozzano, Cimatti, Katoen, Katsaros, Mokos, Nguyen, Noll, Postmac, Roveri. Spacecraft early design validation using formal methods, Reliability Engineering and System Safety, 2014! Mandaras, Early design validation of the GOES I-M system, Master Thesis, AUTh, 2015 Ongoing ESA TRP studies! Catalogue of System & Software Properties with EPFL RiSD Lab and TAS - Requirements catalogue & formalization - Ontology-based semantics modeling & reasoning (Prof. Bassiliades) - Rigorous architecture based design (Prof. Sifakis)! Model-Based Schedulability Analysis for Cached & Multicore Processors (working for CERTH) with Verimag Lab, Cobham Gaisler, Deimos Space
The Model Repair problem! Extension of model checking used for design refinement: Given a system model M and some temporal logic property ϕ, where M does not satisfy ϕ find a new model M such that M satisfies ϕ and the changes in M to derive M are minimal with respect to all such M.! Variants from the bibliography: " with constraints (preserve properties) " with controllable states (repair options)
Applications! Model Repair for incorporating fault tolerance in a distributed algorithm Bonakdarpour, Kulkarni, Abujarad. Symbolic synthesis of masking fault-tolerant programs, 2012! Model Repair for fault recovery in componentbased models Bonakdarpour, Bozga, Goessler. A theory of fault recovery for component-based models, 2011! Model Repair for concurrent programs Attie, Cherri, Al Bab, Sakr, Saklawi. Model and Program Repair via SAT Solving, 2015! Model Repair for probabilistic systems Bartocci, Grosu, Katsaros, Ramakrishnan, Smolka, Model repair for probabilistic systems, 2011 Pathak, Abraham, Jansen, Tacchella, Katoen. A Greedy Approach for the Efficient Repair of Stochastic Models, 2015
Model Repair solutions for probabilistic systems I Bartocci, Grosu, Katsaros, Ramakrishnan, Smolka, Model repair for probabilistic systems, TACAS, 2011 For DTMCs and CTMCs,! using parametric probabilistic model checking the problem is reduced to a nonlinear optimization problem with a minimal-cost objective function! solution feasibility & optimality conditions are provided! an implementation of the solution technique is provided! " # "#$% "#$% "#&$% '%! $ # "#&$%! % # "#$%&' ( %! "#$)' ( % %" '"!" (% &" $" (% "#$%*' #" ( % P!0.3 [F s = 2 " s = 5] # 8v 2 1 + 2 4v 2 1 + 3! 0.3 0/%078-./0+1-%% "#$% (% +%!" $(+'" $(+" $(*'" $(*" $()'" $()" #-%",$('"! " # "#$%+%) ' %! $ # "#&$%+%) & % "#$%(%) ' % '% $" ' ( % "#&$%+%)! %! % # Z "#$%(%*) &% +%)!,%,-./0+01023%4-5067%% P!b [F s = 2 " s = 5] 8v 1 2 + 2 4v 1 2 + 3! b $('"
Model Repair solutions for probabilistic systems II For MDPs, Chen, Hahn, Han, Kwiatkowska, Qu, Zhang, Model Repair for Markov Decision Processes, TASE, 2013! Region refinement through the parameter space (approximation)! Sampling-based search through the parameter space For DTMCs + CTMCs, Pathak, Abraham, Jansen, Tacchella, Katoen. A Greedy Approach for the Efficient Repair of Stochastic Models, 2015! From initial parameter assignment, iteratively changes the parameter values by local repair steps
Abstract Model Repair for transition systems I Remove State d S Add May d 1 s s Remove Must d S α(s) γ(ŝ) α(s ) Remove May d S 2 Change Label d S ŝ Add State d 1 Add Must d S Chatzieleftheriou, Bonakdarpour, Katsaros, Smolka. Abstract Model Repair, NASA Formal Methods 2012 + Logical Methods in Computer Science 2015! Model Repair CTL properties using abstraction & refinement to tackle state space explosion: " Concrete model is a Kripke Structure " Abstract model is a (Kripke) Modal Transition System " A pair of abstraction & concretization functions (a, γ) is defined! A metric space over Kripke structures is defined to quantify their structural differences.! Partial ordering of basic abstract repair operations in terms of the structural changes implied for the concrete model.
Abstract Model Repair for transition systems II Failure Initial Concrete Model (KS) M Abstraction α(m) Abstract Model (KMTS) ˆM = α(m) Refinement Failure Repaired Abstract Model (KMTS) ˆM Abstract Model Repair α Refined (M) No Abstract MC ( ˆM,ŝ) = ϕ? Undefined γ( ˆM ) Concretization Repaired Concrete Model (KS) M Repaired γ( ˆM ) Yes (M,s) = ϕ
Conclusions! Model Repair solutions for probabilistic systems! Abstract Model Repair framework & algorithm " proved sound for the full CTL and complete for a subset of CTL (excluding only the AND operator) " complexity: upper bounded by a polynomial expression in the size of the abstract model " constraints in model repair undermine completeness! Towards Design Repair " better criteria for quantifying changes and minimality (structural differences, only good for abstract repair) " define basic repair operations in rigorous system design languages (e.g. SLIM, BIP) and assess their cost " introduce architecture specific repair options in the design/verification front-end
THANK YOU! katsaros@csd.auth.gr